Upload README.md with huggingface_hub

README.md

@@ -17,8 +17,8 @@ is conditioned on **CLIP-L** text features, with classifier-free guidance.
 
 ## Intended use
 
-Text-to-image generation at 1024×1024. Good at single objects and cinematic scenes; usable single-person
-portraits. A sibling 512px checkpoint additionally does instruction-based image editing.
+Text-to-image generation at 1024×1024. Strongest on single objects and cinematic scenes. A sibling 512px
+checkpoint additionally does instruction-based image editing.
 
 ## How it works
 
@@ -51,10 +51,9 @@ rather than FID / GenEval (which we did not compute):
 | Resolution | 1024×1024 (32×32×32 latent) |
 | VAE reconstruction | ~26 dB PSNR @512px; sharper at 1024px (32×32 latent) |
 
-Qualitatively, the final checkpoint produces **watermark-free**, accurate objects, cinematic scenes, and
-coherent full-body humans. It is **soft on hands and multi-person scenes** — the real small-model /
-latent-resolution ceiling. Loss was still dropping at the end of training, so the 333M DiT is not yet
-saturated.
+Qualitatively, the final checkpoint produces accurate objects and cinematic scenes. It is **soft on people,
+hands, and multi-person scenes** — the real small-model / latent-resolution ceiling. Loss was still dropping
+at the end of training, so the 333M DiT is not yet saturated.
 
 ## Files